June 1, 2009
I develop a theoretical model to explain observed patterns of medical care demand and test the hypothesis that demand is greater the greater the decline in health at any level of health. Medical care demand is highly skewed: the top 5% of individuals consume nearly 50% of expenditures, and nearly half of lifetime expenditures occur after age 65. Extant economic models don't explain this behavior. For example, Murphy & Topel (2006) suggest the willingness to pay for health decreases with age and illness. Grossman (1972) concludes that we demand less health over time, and maintained assumptions about health transition make observed spikes in medical spending unlikely. Tomas Philipson (2007 iHEA plenary) suggested either consumers act irrationally or economists have not adequately modeled behavior. I explore the latter explanation.
I specify an optimal control model that extends the seminal Grossman (1972) model in three ways. I include the change in health in utility; I model depreciation as an amount rather than a rate; and I allow the health state to increase health production. Contrary to the Grossman model, the resulting demand for health suggests an inevitable disequilibrium as health declines between increasing benefits and declining costs of health capital that individuals can only balance by increasing medical care. The time path for medical care demand suggests the change in health rather than the state of health drives increasing demand and that price sensitivity declines over time.
I test the central hypothesis that the change in health is significant using the first 14 waves of the British Household Panel Survey (BHPS). I specify a non-linear seemingly unrelated system of demands for consumption and medical care and impose symmetry restrictions on the cross-price parameters so that inferences are consistent with utility maximization theory. I identify instruments for unobservable health and price using a multiple correspondence analysis. I find support for the theory and the assumption that health and wealth are not separable.
Results suggest single period, single equation models of medical care demand omit relevant variables that capture dynamic decision making and the relationship between health and wealth.